Shoulder mobility exercises, sorted by what they load

Most lists of shoulder mobility exercises are sorted by how the movement looks. This one is sorted by which link of the chain it loads, because the commonest reason weeks of chest and lat stretching change nothing is that the constraint was one segment lower and never got worked.

An arm going overhead needs three things to move: the shoulder in its socket, the shoulder blade across the ribs, and the upper back into extension. Pick the wrong one and you can be diligent for a month and measure the same number.

The short answer

• Sort the work by link: the joint, the blade, the upper back. Doing all three badly beats doing one of them at random.
• The app's catalogue holds nine shoulder moves, named below, plus thoracic moves that do the third job.
• Dose: 30-second holds are enough, and total weekly time is the variable that matters, not the length of any single hold.
• Strength work earned its place here. Overhead is a position you hold as well as reach.
• Measure, work three weeks, measure again — both sides, same conditions.

Shoulder mobility exercises, sorted

The whole shoulder shelf of the catalogue, with what each one is. Nine moves, and the type matters as much as the name.

Two of those carry labels worth knowing. M033 Sleeper Stretch is the catalogue's only level-3 shoulder move, and the app withholds it if you have flagged shoulder impingement; M034 Cross-Body Shoulder Hold is its gentler predecessor. M035 Floor Pec Opener is the only PNF-capable move on this shelf — hold, contract at about 40 per cent effort for six seconds, release for two, then deepen for 15 to 20 seconds, two or three rounds. The app never does that in a warm-up.

Work for the joint itself

These load the ball in the socket and the tissue immediately around it. Use them when the reach behind the back is short and the upper back moves fine.

M036 Standing Shoulder CARs is the one to start with: reach an arm forward at shoulder height and draw the biggest slow circle you control, up, behind and down, turning the palm as it travels. Forty seconds a side, cleared for warm-ups. It is the closest thing in the catalogue to a full survey of the joint's own range, and it tells you where in the arc the trouble sits.

M037 Behind-Back Strap Slide takes the Apley position under control. One hand overhead on a strap, the other behind the lower back, and the hands slide gently closer and apart. Forty seconds a side. It works the exact position the shoulder test reads, which makes it the most direct answer to a low reach score.

M033 Sleeper Stretch targets the back of the shoulder — lie on your side, elbow bent to ninety degrees, and guide the forearm slowly toward the floor. Thirty seconds a side, and stop at the first mild stretch behind the shoulder. M034 Cross-Body Shoulder Hold does a milder version standing.

M005 Doorway Pec Stretch and M035 Floor Pec Opener open the front. Thirty and 45 seconds. They are the two everyone already knows, and they are the two most likely to be the wrong answer, because a short chest is far less common than a stiff upper back.

M038 Overhead Side-Lean loads the lats and the side of the trunk with the arm held high against a wall. Forty seconds a side. Ease off if you feel it pulling in the lower back rather than the side, which means the lumbar spine took the movement.

Work for the shoulder blade

The blade has to rotate upward and tilt back across the ribs for the socket to stay under the arm. Two moves in the catalogue load that directly.

M031 Wall Slides — back, hips and head on a wall, arms in a goalpost, slide overhead keeping the hands near the wall and the ribs down. Dynamic, 45 seconds, cleared for warm-ups, ease off at any pinch in the front of the shoulder. It is also the best unscored self-check on this page: what leaves the wall first tells you which link ran out.

M032 Prone Swimmer Sweep — face down, arms overhead, lift the hands an inch and sweep them out and down toward the hips, drawing the blades together as they travel. Dynamic, 45 seconds, warm-up cleared.

Be careful what you claim for this pair. Scapular control sets out why blade position is a much weaker explanation for symptoms than it is usually made to sound. These are reasonable movements for the region; they are not a correction of anything.

Work for the upper back

This is the link most people are short of and the one that gets least attention. It is not on the shoulder shelf, because the app files it under the thoracic spine.

M041 Kneeling Thoracic Extension — hands well forward on a mat, hips over the knees, let the chest sink toward the floor. A 45-second hold, and one of the moves the app can progress in depth over weeks. M039 Lying Open-Book rotates rather than extends, 45 seconds a side. M044 Supported Backbend Hold does extension over a foam roller, 45 seconds. M001 Cat-Cow is the 45-second dynamic warm-up version.

Thoracic spine mobility covers the segment properly, including the awkward fact that the battery cannot read it: the seated rotation test that would score it is designed and not built, in the app either.

How long and how often

The dose research is unusually clear for a mobility topic, and it is clear in an unglamorous direction.

Bandy and Irion, in Physical Therapy in 1994, compared hold lengths in 57 people: 30 seconds beat 15 and matched 60. Three years later, in the same journal, Bandy, Irion and Briggler found that one 30-second hold a day matched three. Thomas and colleagues, in the International Journal of Sports Medicine in 2018, pooled the trials and concluded that total weekly time is the variable, not how it is divided. Grade: reasonably solid, and taken from hamstrings rather than shoulders — which is where nearly all of this literature lives.

So: 30 to 45 seconds a position, both sides, most days, and stop counting the seconds and start counting the week. Holds run 20 to 60 seconds in the app. Dynamic moves run one repetition per breath cycle, six to twelve of them, with the breath at four seconds in and six out. The target sensation is a mild stretch, never sharp pain.

A session has a shape: one or two dynamic moves as a warm-up, 60 to 90 seconds; then the main block; then one calm long hold to finish, 45 to 60 seconds.

What the evidence says

Four findings, printed in the order of how much they should change what you do.

Strength training moves range about as well as stretching. Afonso and colleagues, in Healthcare in 2021, found comparable range gains from strength training and stretching — on a small and varied set of trials, so treat it as a challenge to the assumption rather than a settled result. Grade: mixed, and the shoulder is where the argument is strongest, because overhead is a position you have to hold as well as reach. A shoulder that can get to a position it cannot control has gained a number and not much else.

Two minutes a day of resistance training reduced neck and shoulder pain in office workers. Andersen and colleagues, in Pain in 2011, and no stretching or mobility routine has a randomised trial with a clinical outcome like it. Grade: moderate, and it is the strongest thing on this page. If discomfort at a desk is what brought you here, that is the better-supported answer.

Exercise for rotator cuff disease has small benefits on low-quality evidence. Page and colleagues' Cochrane review in 2016 found no clear winner among programmes. Grade: weak, in the sense that nobody can tell you which programme. That cuts against every confident routine on the internet, including any version of this one.

Range work is not injury prevention. Lauersen, Bertelsen and Andersen, in the British Journal of Sports Medicine in 2014, found strength training cut injuries to under a third while stretching showed no significant effect. Grade: solid. Stretch a shoulder because you want the range, not because you were promised protection.

Pain that persists, wakes you at night, radiates into a limb, comes with numbness or weakness, or followed a fall or an injury is a matter for a clinician who can examine you. Nothing here diagnoses anything and nothing here is treatment.

Measure, work, re-measure

Do not run any of this blind. The shoulder mobility test takes the Apley scratch in two minutes — five bands, both sides, a 0–100 reading off the app's own anchors. Write the two numbers and the date down. The Apley scratch test explains what those bands can and cannot tell you, including why a one-band difference between your sides means measure again rather than train harder.

Then work for three weeks and read it again under the same conditions: same hour, same warm-up, same naming of sides. Three weeks is the app's default gap, and a reading over 21 days old is marked stale rather than treated as current. The Range Score composes the reading with the rest of the battery.

If the number has not moved, change one thing — usually which link you were loading — rather than adding volume to the link that was never the problem. The rest of the cluster is shoulders and upper back.

Questions

How long should I hold a shoulder stretch?

Thirty seconds is enough for a single hold, and the trials that established it found 60 no better and 15 worse. What matters more is the total across the week, so most days at 30 to 45 seconds a position beats one long weekend session. The app's holds run 20 to 60 seconds.

Which move should I start with if my reach is short?

M037 Behind-Back Strap Slide, because it works the exact position the shoulder test reads, and M036 Standing Shoulder CARs, because a slow controlled circle shows you where in the arc the restriction sits. Add M031 Wall Slides for the blade and the upper back. Chest stretches are the popular starting point and the one most likely to be aimed at the wrong link.

Is stretching or strengthening better for shoulder range?

The comparison has been made and it came out close: a 2021 review found comparable range gains from both, on a small and varied set of trials. For the shoulder specifically the strength side has an extra argument, because overhead is a position you have to hold under load as well as reach. The clearest clinical result in this area is for a small daily dose of resistance training in office workers.

How often should I do this?

Most days, briefly, rather than twice a week at length. Range responds to accumulated weekly time, and gains reverse when the work stops, so a routine you keep beats a better routine you abandon. Two to five minutes on the shoulders inside a longer session is a reasonable dose.

Will these movements help a painful shoulder?

That is not a claim this page makes. Exercise-based care is what most guidelines recommend for rotator cuff related shoulder pain, but the Cochrane review found small benefits on low-quality evidence and no clear winner among programmes. If a shoulder is painful — especially at night, with weakness, or after an injury — see a clinician who can examine it rather than picking a routine off a web page.

How do I know any of it worked?

Take a scored reading before you start and another three weeks later, in the same conditions, with both sides recorded separately. A band change on the shoulder test is a real change; a nicer feeling after a session is not, because acute range gains have been shown to fade within a few minutes. Same hour, same warm-up, date written down.